Buyer Behavior Models and Attribute Models: a Synthesis


Donald R. Lehmann (1972) ,"Buyer Behavior Models and Attribute Models: a Synthesis", in SV - Proceedings of the Third Annual Conference of the Association for Consumer Research, eds. M. Venkatesan, Chicago, IL : Association for Consumer Research, Pages: 526-535.

Proceedings of the Third Annual Conference of the Association for Consumer Research, 1972      Pages 526-535


Donald R. Lehmann, Columbia University


Recently there has been substantial interest in developing models of individual consumer choice. Two of the most popular types of models are large-scale models of buyer behavior of the type proposed by Nicosia (1966) and Howard and Sheth (1969) and attribute models of preference based on the models of Fishbein (1967) and Rosenberg (1956). This paper will begin by summarizing the current state of research in these two areas, and then will suggest how the two types of models might be combined.


Buyer behavior models have addressed the question of how a buyer goes about gathering information for making a decision, how he makes a decision, and finally how the decision affects his attitudes and hence future decisions. In other words, they are attempts to describe buyers from "cradle to grave." These models are thus directed at the Herculean task of explaining buyer behavior in every facet.

These buyer behavior models are usually stated in terms of a flowchart. These flowcharts suggest the general direction of flows from one endogenous variable to another. They do not, however, provide operational definitions of the constructs in each box of the flowchart, nor do they in general specify what exogenous variables affect the various endogenous variables. Moreover, they do not specify the mathematical form of the links between variables. Thus as such, these flowcharts are difficult to operationalize and study empirically (and in a predictive testing sense, impossible to test at their current stage of development).

Because of the problems involved in investigating these models, most initial "tests" of the models have been relatively simplistic, albeit relatively sophisticated statistically. Using examinations of the Howard-Sheth model as an example (Farley & Ring, 1970; Lehmann, Farley, & Howard, 1971; and Lehmann, O'Brien, Farley & Howard, 1971), several interesting observations are possible:

1. The "tests" have been largely cross-sectional.

2. The mathematical form used has been linear.

3. Parameter estimates have been made across people using regression (either OLS or TSLS).

4. The results are encouraging but mixed. The links in the cognitive side of the model, including such variables as brand comprehension, attitude, intention, and purchase have been both significant and plausible. The links among the informational variables, such as attention, perceptual bias, and overt search, on the other hand, have been much weaker.

At this point, many areas of the Howard-Sheth model are largely unexplored, including:

1. Non-linear forms

2. Lagged forms

3. Alternative operational definitions 4. Individual parameter estimates.

Thus the major characteristic of these general buyer behavior models is their limited operationalization.


Perceptual mapping models differ substantially from full-scale buyer behavior models. They focus on explanation of individual preference, and are not immediately concerned with either information reception on the one hand or choice on the other. The essential feature of these models is that they view brands as a collection of positions on a set of attributes, and preference toward a brand as some weighted combination of the positions.

Perceptual mapping models can be expressed graphically as in Figure 2. The essential postulate of these models is that the "closer" an alternative is to the ideal, the more preferred it is. In Figure 2, this would imply, assuming attributes 1 and 2 were equally important. that alternative A is the most preferred.

A variety of trends in the literature has suggested the perceptual mapping approach. In economics, Lancaster (1966) has proposed a utility theory based on the characteristics of a good instead of the good as a whole. The multidimensional scaling literature suggests that preference is a function of the distance of an object from the ideal (Green & Carmone, 1969; Kruskal, 1964 a & b; and Shepard, 1962 a & b). In social psychology, two very similar theories of attitude have been provided (Fishbein, 1967; and Rosenberg, 1956) which suggest that attitude is a weighted sum of positions on dimensions. Thus the essential concept that an object can be viewed as a point in n dimensional space is widely supported.

One reason why these research traditions were not merged sooner is the differences in terminology used to describe them. In order to make the similarity more obvious, the following glossary is useful:


With one important exception (Einhorn & Gonedes, 1971), all the perceptual mapping models proposed have been of the following form:



Wi = weight of the ith dimension

Pji = position of the jth object on the ith dimension

Ii = ideal position on the ith dimension

K = an integer

and n = number of relevant dimensions.

In other words, the attitude is a weighted sum of distance to the ideal on each of the relevant dimensions.

Depending on the way Ii and K are defined, Yj can take on many forms. For example, a K = 1 implies city block distance while a K = 2 implies Euclidean distance. In two past tests, city block distance has proven best predictively (Bass, Pessemier & Lehmann, 1971; and Lehmann, 1971). However, other considerations, such as stability under orthogonal rotation (city block distance is not, while Euclidean is) and utility theory implications (city block implicitly assumes constant marginal utility on the attributes, while Euclidean is one form of diminishing marginal utility) may dominate in the selection of a distance measure. In any event, several alternative distance measures are available.

The relationship of distance to the ideal, similarity (which is the inverse of distance) to the ideal, attitude, and preference are also somewhat confusing. The relationship can be summarized as follows:


Actually, preference is usually a comparative measure between attitudes, but for purposes of these perceptual mapping models, the two terms are largely synonymous. Thus the differences between the traditions in the literature are largely semantic.

A more fundamental reason why these traditions in the literature were not synthesized sooner is that there are essential differences between the models in terms of the way the dimensions arise. Two basic approaches exist, and they differ in some important features:


Yet in spite of these differences, the essential similarity of these approaches is obvious.

Tests of these perceptual mapping models have differed substantially from those of the Howard-Sheth model. Because the mathematical form of the relation is pre-specified, the model is more operational. Efforts have centered around measurement of the three basic constructs (preference, weight, and position) and deducing the relevant dimensions.

In general, the tests have been greatly encouraging. Indirectly derived dimensions have proved useful in explaining preference among such diverse alternatives as automobiles (Green & Carmone, 1969), jobs (Hill & Pessemier, 1971), and political candidates (Johnson, 1970). Ratings on pre-specified dimensions have proved successful in analyzing such alternatives as television shows (Lehmann, 1971), soft drinks (Bass, Pessemier & Lehmann, 1971), and numerous branded products (Bass & Talarzyk, 1972; Ginter, 1972; and Winter, 1972). In all these examples, predictions based on a perceptual mapping model have greatly outperformed both demographics and random models.

In spite of these encouraging results, there are some important problems involved with applying perceptual mapping models. As stated, these models are largely tautological and as such can be investigated but not truly tested. Also attempts to use a subject-estimated ideal point have been disappointing (Bass et al; 1971; and Lehmann, 1971). Finally, the weights have not proved to be very useful (Lehmann, 1971; and Sheth & Talarzyk, 1972) for a variety of reasons (Beckwith & Lehmann, 1972). Thus substantial testing and refinement of these models is also needed.


Looking at the pictorial representations of the two models of individual behavior (Figures 1 and 2), one is struck more by the differences than by the similarities. Considering the problems involved in testing either separately, the obvious question which arises is: "Why attempt to synthesize them?" The answer is that by combining them, both may benefit.

The obvious weakness of the perceptual mapping approach is that it does not suggest either how information influences the individual or how preference is related to choice. Placing it in the context of a general buyer behavior model suggests both. On the other hand, the obvious weakness of the buyer behavior approach is the imprecise formulation of the links between blocks in the flowchart. Using perceptual mapping makes some of the links both explicit mathematically and empirically viable.

To see how a perceptual mapping model might be combined, consider the Howard-Sheth model (Howard & Sheth, 1971), which is currently under revision (Howard & Ostlund, 1973), portrayed in Figure 1, and the perceptual mapping model represented by Figure 2. The perceptual mapping model can be viewed as a combination of four constructs: 1) Choice criteria, 2) Weights, 3) Brand comprehension, and 4) Confidence. The choice criteria can be viewed as the dimensions of the perceptual map and the weights as the weights attached to the dimensions. Brand comprehension could be treated as the position of the brand on the dimensions. Finally, the random component representing uncertainty suggested for introduction into the perceptual mapping models (Lehmann, 1971 a;$ 1972 a $ b) can be considered as a measure of confidence. In other words, the center of the Howard-Sheth model could be viewed as a perceptual mapping model.





One advantage of this synthesis is the improved explanatory power for the center of the Howard-Sheth model. Another is that the "new" model can be expressed very simply (Figure 3). Possibly the most important advantage of the synthesis, however, is that it suggests how advertising and other information affects choice. Simply stated, this new model suggests that advertising influences choice by changing either the positions on the dimensions, the weights of the dimensions, or the ideal position on the dimensions. As such, it provides a structure for future research on the informational side of the model which has to date proved the most difficult to investigate.


Two major research traditions dealing with the individual consumer have emerged: Buyer Behavior Models and Perceptual Mapping. In spite of the problems involved in attempting to test (or at least investigate) these models, both have shown great promise. This paper suggests that combining the two traditions will be to the mutual benefit of both.




Bass, Frank M., Edgar A. Pessemier, & Donald R. Lehmann. An Experimental Study of Relationships Between Attitudes, Brand Preference, and Choice. Paper No. 307, Institute for Research in the Behavioral, Economic, and Management Sciences, Herman C. Krannert Graduate School of Industrial Administration, Purdue University, April, 1971.

Bass, Frank M. & W. Wayne Talarzyk. An Attitude Model for the Study of Brand Preference. Journal of Marketing Research, 1972, 9, 93-96.

Beckwith, Neil E. & Donald R. Lehmann. The Importance of Importances in Attribute Models of Consumer Preference. Working paper, Columbia University Graduate School of Business, 1972.

Einhorn, Hillel J. & Nicholas J. Gonedes. An Exponential Discrepancy Model for Attitude Evaluation. Behavioral Science, 1971, 16, 152-157.

Farley, John U. & L. Winston Ring. An Empirical Test of the Howard-Sheth Model of Buyer Behavior, Journal of Marketing Research, 1970, 7, 427-438.

Fishbein, Martin. Attitude and the Prediction of Behavior. In Martin Fishbein (Ed.), Readings in Attitude Theory and Measurement. New York: John Wiley & Sons, 1967, pp. 477-492.

Ginter, James L. An Experimental Study of Attitude Change and Choice of New Brands. Proceedings. Spring Conference, American Marketing Association, 1972.

Green, Paul E. & Frank J. Carmone. Multidimensional Scaling: An Introduction and Comparison of Nonmetric Unfolding Techniques. Journal of Marketing Research, 1969, 4, 330-341.

Hill, Raymond E. & Edgar A. Pessemier. Multidimensional and Unidimensional Metric Scaling of Preference for Job Descriptions. Paper No. 308, Institute for Research in the Behavioral,Economic, and Management Sciences, Herman C. Krannert Graduate School of Industrial Administration, Purdue University, April, 1971.

Howard, John A. & Lyman E. Ostlund. Buyer Behavior: Theoretical and Empirical Foundations. New York: Alfred A. Knopf, Inc., 1973 (in press).

Howard, John A. & Jagdish N. Sheth. The Theory of Buyer Behavior. New York: John Wiley & Sons, 1969.

Johnson, Richard M. Political Segmentation. Marketing Review, 1970, 25, 20-24.

Kruskal, J. B. Multidimensional Scaling by Optimizing Goodness of Fit to a Nonmetric Hypothesis. Psychometrika, 1964a, 29, 1-27.

Kruskal, J. B. Nonmetric Multidimensional Scaling: A Numerical Method. Psychometrika, 1964b, 29, 115-129.

Lancaster, Kelvin J. A New Approach to Consumer Theory. Journal of Political Economy, 1966, 74, 132-157.

Lehmann, Donald R. A Study of Television Show Preference and Attributes of Programs. Proceedings. Spring Conference, American Marketing Association, 1972a.

Lehmann, Donald R. Preference Among Similar Alternatives. Working paper, Columbia University, 1972b.

Lehmann, Donald R. Evaluating Marketing Strategy in a Multiple Brand Market. Journal of Business Administration, 1971a, 3, 15-26.

Lehmann, Donald R. Television Show Preference: Application of a Choice Model. Journal of Marketing Research, 1971b, 8, 47-55.

Lehmann, Donald R. John U. Farley $ John A. Howard. Testing of Buyer Behavior Models. Proceedings. Second Annual Conference, Association for Consumer Research, SePtember 1-3, 1971c, 232-242.

Lehmann, Donald R. Terrence V. O'Brien, John U. Farley, & John A. Howard. Empirical Contributions to Buyer Behavior Theory. Working paper, Columbia UniversitY, 1971.

Nicosia, Francesco. Consumer Decision Processes: Marketing and Advertising Implications. Englewood Cliffs, N.J.: Prentice-Hall, 1966.

Rosenberg, M. J. Cognitive Structure and Attitudinal Affect. Journal of Abnormal and Social Psychology, 1956, 53, 367-372.

Shepard, Roger N. The Analysis of Proximities: Multidimensional Scaling with an Unknown Distance Function L Psychometrika, 1962a, 27, 125-139.

Shepard, Roger N. The Analysis of Proximities: Multidimensional Scaling with an Unknown Distance Function II. Psychometrika, 1962, 27, 219-246.

Sheth, Jagdish N. $ W. Wayne Talarzyk. Perceived Instrumentality and Value Importance as Determinants of Attitudes. Journal of Marketing Research, 1972, 9, 6-9.

Winter, Frederick W. Laboratory Experimental Study of Attitude Change and Brand Choice. Proceedings. Spring Conference, American Marketing Association, 1972.



Donald R. Lehmann, Columbia University


SV - Proceedings of the Third Annual Conference of the Association for Consumer Research | 1972

Share Proceeding

Featured papers

See More


Self-Deprecation Signals Humility, but Not as Much as Self-Deprecators Assume

Clayton R Critcher, University of California Berkeley, USA
Michael O'Donnell, University of California Berkeley, USA
Minah Jung, New York University, USA

Read More


F9. Protection against Deception: The Moderating Effects of Knowledge Calibration on Consumer Responses to Ambiguous Advertisement Information

Joel Alan Mohr, Queens University, Canada
Peter A. Dacin, Queens University, Canada

Read More


H6. The Influence of Anthropomorphized Service Toolkit on Consumer Satisfaction in Service Coproduction

Rocky Peng Chen, Hong Kong Baptist University
Kimmy Chan, Hong Kong Baptist University

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.